Color 2/19
correctly colored objects are identified more accurately and quickly than incorrectly colored ones
how do we see color?
physics of light interacted with our perceptual mechanisms and the brain to determine the colors we see
our perception of color is a figment of our brains
physics:
white light (light from the sun) contains many wavelengths
Isaac Newton:
was the prism creating colors or splitting white into basic elements?
why does a prism separate colors in white light?
when light travels through a medium, it bends toward the widest part of the medium
the amount of bending is related to how many cycles occur within the medium
shorter wavelengths will have more cycles in the same amount of space, so short will curve more
Newton and Spectral colors:
7 spectral colors- ROYGBIV (or 6, some people no longer say indigo)
all other colors are non-spectral
no single wavelength will make them
they exist because our brain interprets combinations of spectral colors as unique colors (ex. green + blue = cyan)
why does an apple look red?
selectively reflects red
then apple absorbs short wavelength light and reflects long wavelength light (physics)
if you measure the light coming off the apple it will have mostly long wavelengths
spectral reflectance: object’s surface reflects more or less of the light at different wavelengths
reflectance curve of an object plots of percentage of light reflected for specific wavelengths
Trichromatic Theory:
Thomas Young- instrumental in deciphering Rosetta Stone, British Polymath, 1773-1829
Hermann Von Helmholtz- German physicist
their evidence: color matching task
observers were presented with two screens separated by a partition
task was to manipulate the intensity of three different, superimposed colored light to match the test light on the other side of the partition
participants with normal color vision, three spotlights (red, green blue) are enough to match any color of test light
trichromatic theory posits that the neural representation of color reduces color to values of only three color channels-
the amount of red, green, and blue
patterns of responses across these three channels allow for the perception of all color
the logic: if any color can be made by adjusting amount of RGB light in the stimulus, the nervous system must code the amount of RGB to see all colors
now: we have identified that there are three unique cones
each w different opsin (diff photopigment) w different absorption spectrum
short, medium, long cones
corresponding to preferred wavelength, not size
opponent process theory: Ewald Hering- German physiologist who contributed a lot to color vision, hyper acuity, and binocular vision
three color mechanisms operate in opposing fashion to all perception of color
red/green
yellow/blue
black/white
trichromacy accounts for the response of the photoreceptors in the retina
opponent process accounts for the response of retinal ganglion cells connected to the cones
metamers: two stimuli that appear similar but are physically very different- the perceived matching of colors that have different spectral power distributions
McCollough effect:
adapt very specific cells (tuned to orientation, width, color), the after effect can last a really long time
suggests after effects are not fatigue, but changes in sensitivity
to rebalance sensitivity cells must be active- specialized cells are rarely active
why can we see so many colors?
only 3 types of cones
each cell can only fire more or less
we can see so many shapes because distributed coding - 3 cones w overlapping response curves allow precise coding of the wavelength
why have oppenency?
to deal with changes in base rates of firing
to make system more sensitive
think Weber’s law- deviations from near zero are easier to detect than deviation from near 10
how we describe color:
hue - wavelength of light
saturation - how pure (less more white light the less saturated), desaturated is lots of white light
brightness - strength of the signal (value in the image)
color mixing: additive vs subtractive
additive means you shine more lights on the same surface
primary colors- red, green, blue
secondary colors (combo of two primaries)- yellow, magenta, cyan
complimentary colors- when added together yield white- 2 colors contain all 3 primaries
red + cyan, green + magenta, blue + yellow
additive color mixture: mixing lights of diff wavelengths, result is symmetrical of all component wavelengths - superimposing blue and yellow lights leads to white
subtractive color mixing- starting with white light (all colors) and taking some away - pigments selectively absorb wavelengths
subtractive color mixture:
mixing paints w different pigments
mixture reflects wavelengths that are reflected in common by components
mixing blue and yellow leads to green
primaries - allows you to selectively alter a single cone type (turn it off rather than on) (look at page 10, slide 55 of color lecture)
key to understanding additive vs. subtractive:
in additive start in a dark room and turn lights on
in subtractive start with white lights (all wavelengths) and pigments absorb them
color mixing real world applications:
RGB- TVs- additive mixing
CMYK- printing- cyan, magenta, yellow, black- subtractive mixing of absorbing pigments
optical color mixing: pointillism
mixture created is more vivid than when mixing paints because one does not remove light
lightness constancy:
perceiving the surface to have the same lightness under illumination of different amounts of lights
in dim (indoor) illumination white may only reflect 90 units of light
same checkerboard outside, black may reflect 900 units of light, so black is 10x brighter outside than white is inside
Adelson’s checker-shadow illusion
brain automatically accounts for perceived changes in illumination (ex. shadows) discounting the illuminant
shadows:
not parsed as objects- tend to be ignored
tell us about how to discount the illuminant, can tell us about objects
color constancy: an extension of lightness constancy
spectral reflectance: object surfaces reflect different amount of light depending on the wavelengths
reflectance curve of an object plots of percentage of light reflected for specific wavelengths
illuminant:
sunlight has roughly equal light from across the spectrum
light bulbs can have different spectrums
Retinex Theory (Edwin Land):
assume a white object is the brightest thing in the environment and reflects all wavelengths at the same rate
Possible mechanisms of color constancy:
discount the illuminant
estimate the spectral power distribution of the illuminant
precise mechanism not clear (Retinex theory based on anchoring theory = an approximation)
chromatic adaptation
prolonged exposure to a color leads to adaptation
if the illuminant has too much red - the constant stimulation of red will cause a decrease in the red response
this will have the consequence of “turning down” the red in the perception of the scene (which is similar to discounting the illuminant)
mostly some form of anchoring (retinex theory)
some adaptation
a test of chromatic adaption account:
observers shown sheets of colored paper in three conditions-
baseline- paper and observer in white light
un adapted- paper illuminated by red light; observer by white light
adapted- paper and observer in red light
results
baseline - green paper seen as green
unadapted- perception of green paper is shifted toward red (not complete color constancy)
adapted- perception of green paper is slightly shifted toward red (partial color constancy)
The Dress:
two possible interpretations:
the dress is outside but in a shadow
to discount the shadow your brain lightens image and removes some blue (shadows darken and emphasize short/blue light)
result: gold and white
the dress is inside and light by artificial light (too much yellow)
to discount the illuminant your brain darkens and removes yellow
result: bluish becomes more blue; darker gold becomes black
color constancy:
anchoring theories (like Retinex) require broadband illumination
need to have the full spectral curve to color correct
in support of anchoring theories - color constancy breaks down under spectral illumination
low pressure sodium lamp: emits monochromatic light, color constancy breaks down, everything looks grey
color anomalous vision (color blindness):
dichromats - missing 1 of the 3 photopigments
most common - protanope (red cone cells defective) and deuteranope (green cone cells defective) (have both cone types but same opsin in both)
tritanope (blue cone cells defective)
if a colorblind woman has a son the odds he is color blind is 100%
color deficient experiences is predictable from opposition coding
if “green” and “red” cones have the same opsin, they will always respond equally
so in red-green channel signal will always cancel, but will drive the yellow channel
testing “color blindness” - Ishihara color plates
Dogs vs Humans
dog fovea - 20% cones, 80% rods
dogs are dichromats- 2 cone types
humans 3 cones have peak sensitivities of 419, 531, 558nm
dogs 2 cones have peal sensitivities of 429 nm and 555
Cortical Processing: area V4
V4- long thought to be cortical area associated with conscious color perception
cells in V4 have color selectivity (and larger RFs)
cortical color blindness (cerebral achromatopsia)
damage to V4 procures loss of color experience
conclusion:
perception of color requires:
light energy from the visible spectrum
analysis of the photoreceptor’s activations
cortical derivation of color (perception)
“color” is a mental science governed by the laws of our brains